1999-10-28 21:36:25 +05:30
|
|
|
/*
|
2000-07-11 04:25:51 +05:30
|
|
|
* sed.c - very minimalist version of sed
|
1999-10-28 21:36:25 +05:30
|
|
|
*
|
2000-04-13 06:48:56 +05:30
|
|
|
* Copyright (C) 1999,2000 by Lineo, inc.
|
2000-07-11 04:25:51 +05:30
|
|
|
* Written by Mark Whitley <markw@lineo.com>, <markw@enol.com>
|
1999-12-30 03:49:46 +05:30
|
|
|
*
|
1999-10-28 21:36:25 +05:30
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/*
|
|
|
|
Supported features and commands in this version of sed:
|
|
|
|
|
|
|
|
- comments ('#')
|
2000-07-14 05:30:15 +05:30
|
|
|
- address matching: num|/matchstr/[,num|/matchstr/|$]command
|
|
|
|
- commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags)
|
|
|
|
- edit commands: (a)ppend, (i)nsert, (c)hange
|
2000-07-18 01:36:42 +05:30
|
|
|
- backreferences in substitution expressions (\1, \2...\9)
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
(Note: Specifying an address (range) to match is *optional*; commands
|
|
|
|
default to the whole pattern space if no specific address match was
|
|
|
|
requested.)
|
|
|
|
|
|
|
|
Unsupported features:
|
|
|
|
|
|
|
|
- transliteration (y/source-chars/dest-chars/) (use 'tr')
|
|
|
|
- no support for characters other than the '/' character for regex matches
|
|
|
|
- no pattern space hold space storing / swapping (x, etc.)
|
|
|
|
- no labels / branching (: label, b, t, and friends)
|
|
|
|
- and lots, lots more.
|
|
|
|
|
|
|
|
*/
|
|
|
|
|
1999-10-28 21:36:25 +05:30
|
|
|
#include <stdio.h>
|
2000-07-11 04:25:51 +05:30
|
|
|
#include <unistd.h> /* for getopt() */
|
|
|
|
#include <regex.h>
|
|
|
|
#include <string.h> /* for strdup() */
|
1999-10-28 21:36:25 +05:30
|
|
|
#include <errno.h>
|
2000-07-11 04:25:51 +05:30
|
|
|
#include <ctype.h> /* for isspace() */
|
2000-09-26 03:15:58 +05:30
|
|
|
#include "busybox.h"
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
/* externs */
|
|
|
|
extern int optind; /* in unistd.h */
|
|
|
|
extern char *optarg; /* ditto */
|
|
|
|
|
|
|
|
/* options */
|
|
|
|
static int be_quiet = 0;
|
|
|
|
|
2000-08-22 02:59:20 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
struct sed_cmd {
|
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
|
|
|
|
/* GENERAL FIELDS */
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* address storage */
|
|
|
|
int beg_line; /* 'sed 1p' 0 == no begining line, apply commands to all lines */
|
|
|
|
int end_line; /* 'sed 1,3p' 0 == no end line, use only beginning. -1 == $ */
|
|
|
|
regex_t *beg_match; /* sed -e '/match/cmd' */
|
|
|
|
regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */
|
|
|
|
|
|
|
|
/* the command */
|
|
|
|
char cmd; /* p,d,s (add more at your leisure :-) */
|
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
|
|
|
|
/* SUBSTITUTION COMMAND SPECIFIC FIELDS */
|
|
|
|
|
|
|
|
/* sed -e 's/sub_match/replace/' */
|
|
|
|
regex_t *sub_match;
|
|
|
|
char *replace;
|
2000-07-18 01:36:42 +05:30
|
|
|
unsigned int num_backrefs:4; /* how many back references (\1..\9) */
|
|
|
|
/* Note: GNU/POSIX sed does not save more than nine backrefs, so
|
|
|
|
* we only use 4 bits to hold the number */
|
2000-11-04 01:17:00 +05:30
|
|
|
unsigned int sub_g:1; /* sed -e 's/foo/bar/g' (global) */
|
|
|
|
unsigned int sub_p:2; /* sed -e 's/foo/bar/p' (print substitution) */
|
|
|
|
|
|
|
|
|
|
|
|
/* EDIT COMMAND (a,i,c) SPEICIFIC FIELDS */
|
2000-07-14 05:30:15 +05:30
|
|
|
|
|
|
|
char *editline;
|
2000-07-11 04:25:51 +05:30
|
|
|
};
|
|
|
|
|
|
|
|
/* globals */
|
|
|
|
static struct sed_cmd *sed_cmds = NULL; /* growable arrary holding a sequence of sed cmds */
|
|
|
|
static int ncmds = 0; /* number of sed commands */
|
|
|
|
|
|
|
|
/*static char *cur_file = NULL;*/ /* file currently being processed XXX: do I need this? */
|
1999-10-28 21:36:25 +05:30
|
|
|
|
2000-07-25 23:31:20 +05:30
|
|
|
#ifdef BB_FEATURE_CLEAN_UP
|
2000-07-11 04:25:51 +05:30
|
|
|
static void destroy_cmd_strs()
|
|
|
|
{
|
|
|
|
if (sed_cmds == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* destroy all the elements in the array */
|
|
|
|
while (--ncmds >= 0) {
|
2000-02-09 01:28:47 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
if (sed_cmds[ncmds].beg_match) {
|
|
|
|
regfree(sed_cmds[ncmds].beg_match);
|
|
|
|
free(sed_cmds[ncmds].beg_match);
|
|
|
|
}
|
|
|
|
if (sed_cmds[ncmds].end_match) {
|
|
|
|
regfree(sed_cmds[ncmds].end_match);
|
|
|
|
free(sed_cmds[ncmds].end_match);
|
|
|
|
}
|
|
|
|
if (sed_cmds[ncmds].sub_match) {
|
|
|
|
regfree(sed_cmds[ncmds].sub_match);
|
|
|
|
free(sed_cmds[ncmds].sub_match);
|
|
|
|
}
|
|
|
|
if (sed_cmds[ncmds].replace)
|
|
|
|
free(sed_cmds[ncmds].replace);
|
2000-02-09 01:28:47 +05:30
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
/* destroy the array */
|
|
|
|
free(sed_cmds);
|
|
|
|
sed_cmds = NULL;
|
|
|
|
}
|
2000-07-13 05:05:21 +05:30
|
|
|
#endif
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
#if 0
|
2000-07-11 04:25:51 +05:30
|
|
|
/*
|
|
|
|
* trim_str - trims leading and trailing space from a string
|
|
|
|
*
|
|
|
|
* Note: This returns a malloc'ed string so you must store and free it
|
|
|
|
* XXX: This should be in the utility.c file.
|
2000-07-15 00:36:30 +05:30
|
|
|
* XXX: This is now obsolete. Maybe it belongs nowhere.
|
2000-07-11 04:25:51 +05:30
|
|
|
*/
|
|
|
|
static char *trim_str(const char *str)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
char *retstr = strdup(str);
|
|
|
|
|
|
|
|
/* trim leading whitespace */
|
|
|
|
memmove(retstr, &retstr[strspn(retstr, " \n\t\v")], strlen(retstr));
|
|
|
|
|
|
|
|
/* trim trailing whitespace */
|
|
|
|
i = strlen(retstr) - 1;
|
|
|
|
while (isspace(retstr[i]))
|
|
|
|
i--;
|
|
|
|
retstr[++i] = 0;
|
|
|
|
|
|
|
|
/* Aside:
|
|
|
|
*
|
|
|
|
* you know, a strrspn() would really be nice cuz then we could say:
|
|
|
|
*
|
2000-07-15 00:36:30 +05:30
|
|
|
* retstr[strrspn(retstr, " \n\t\v") + 1] = 0;
|
2000-07-11 04:25:51 +05:30
|
|
|
*/
|
|
|
|
|
|
|
|
return retstr;
|
|
|
|
}
|
2000-07-15 00:36:30 +05:30
|
|
|
#endif
|
|
|
|
|
|
|
|
#if 0
|
|
|
|
/*
|
|
|
|
* strrspn - works just like strspn() but goes from right to left instead of
|
|
|
|
* left to right
|
|
|
|
*/
|
|
|
|
static size_t strrspn(const char *s, const char *accept)
|
|
|
|
{
|
|
|
|
size_t i = strlen(s);
|
|
|
|
|
|
|
|
while (strchr(accept, s[--i]))
|
|
|
|
;
|
|
|
|
|
|
|
|
return i;
|
|
|
|
}
|
|
|
|
#endif
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
/*
|
2000-07-18 01:36:42 +05:30
|
|
|
* index_of_next_unescaped_slash - walks left to right through a string
|
|
|
|
* beginning at a specified index and returns the index of the next forward
|
|
|
|
* slash ('/') not preceeded by a backslash ('\').
|
2000-07-11 04:25:51 +05:30
|
|
|
*/
|
2000-07-14 01:28:04 +05:30
|
|
|
static int index_of_next_unescaped_slash(const char *str, int idx)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
2000-07-18 01:36:42 +05:30
|
|
|
for ( ; str[idx]; idx++) {
|
|
|
|
if (str[idx] == '/' && str[idx-1] != '\\')
|
|
|
|
return idx;
|
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2000-07-18 01:36:42 +05:30
|
|
|
/* if we make it to here, we've hit the end of the string */
|
|
|
|
return -1;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* returns the index in the string just past where the address ends.
|
|
|
|
*/
|
|
|
|
static int get_address(const char *str, int *line, regex_t **regex)
|
1999-12-30 03:49:46 +05:30
|
|
|
{
|
2000-07-11 04:25:51 +05:30
|
|
|
char *my_str = strdup(str);
|
|
|
|
int idx = 0;
|
|
|
|
|
|
|
|
if (isdigit(my_str[idx])) {
|
|
|
|
do {
|
|
|
|
idx++;
|
|
|
|
} while (isdigit(my_str[idx]));
|
|
|
|
my_str[idx] = 0;
|
|
|
|
*line = atoi(my_str);
|
|
|
|
}
|
|
|
|
else if (my_str[idx] == '$') {
|
|
|
|
*line = -1;
|
|
|
|
idx++;
|
|
|
|
}
|
|
|
|
else if (my_str[idx] == '/') {
|
2000-07-18 01:36:42 +05:30
|
|
|
idx = index_of_next_unescaped_slash(my_str, ++idx);
|
2000-07-11 22:23:56 +05:30
|
|
|
if (idx == -1)
|
2000-07-12 22:32:35 +05:30
|
|
|
fatalError("unterminated match expression\n");
|
2000-07-11 22:23:56 +05:30
|
|
|
my_str[idx] = '\0';
|
|
|
|
*regex = (regex_t *)xmalloc(sizeof(regex_t));
|
2000-11-04 01:53:49 +05:30
|
|
|
xregcomp(*regex, my_str+1, 0);
|
2000-07-14 04:22:02 +05:30
|
|
|
idx++; /* so it points to the next character after the last '/' */
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
|
|
|
else {
|
2000-07-14 07:21:25 +05:30
|
|
|
errorMsg("get_address: no address found in string\n"
|
|
|
|
"\t(you probably didn't check the string you passed me)\n");
|
2000-07-11 04:25:51 +05:30
|
|
|
idx = -1;
|
|
|
|
}
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
free(my_str);
|
|
|
|
return idx;
|
|
|
|
}
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2000-07-11 22:23:56 +05:30
|
|
|
static char *strdup_substr(const char *str, int start, int end)
|
|
|
|
{
|
|
|
|
int size = end - start + 1;
|
|
|
|
char *newstr = xmalloc(size);
|
|
|
|
memcpy(newstr, str+start, size-1);
|
|
|
|
newstr[size-1] = '\0';
|
|
|
|
return newstr;
|
|
|
|
}
|
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
static int parse_subst_cmd(struct sed_cmd *sed_cmd, const char *substr)
|
2000-07-14 01:28:04 +05:30
|
|
|
{
|
|
|
|
int oldidx, cflags = REG_NEWLINE;
|
|
|
|
char *match;
|
|
|
|
int idx = 0;
|
2000-07-18 01:36:42 +05:30
|
|
|
int j;
|
2000-07-14 01:28:04 +05:30
|
|
|
|
|
|
|
/*
|
|
|
|
* the string that gets passed to this function should look like this:
|
2000-08-22 02:59:20 +05:30
|
|
|
* s/match/replace/gIp
|
|
|
|
* || | |||
|
2000-07-14 01:28:04 +05:30
|
|
|
* mandatory optional
|
|
|
|
*
|
|
|
|
* (all three of the '/' slashes are mandatory)
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* verify that the 's' is followed by a 'slash' */
|
|
|
|
if (substr[++idx] != '/')
|
|
|
|
fatalError("bad format in substitution expression\n");
|
|
|
|
|
|
|
|
/* save the match string */
|
|
|
|
oldidx = idx+1;
|
2000-07-18 01:36:42 +05:30
|
|
|
idx = index_of_next_unescaped_slash(substr, ++idx);
|
2000-07-14 01:28:04 +05:30
|
|
|
if (idx == -1)
|
|
|
|
fatalError("bad format in substitution expression\n");
|
|
|
|
match = strdup_substr(substr, oldidx, idx);
|
|
|
|
|
2000-07-18 01:36:42 +05:30
|
|
|
/* determine the number of back references in the match string */
|
|
|
|
/* Note: we compute this here rather than in the do_subst_command()
|
|
|
|
* function to save processor time, at the expense of a little more memory
|
|
|
|
* (4 bits) per sed_cmd */
|
|
|
|
|
|
|
|
/* sed_cmd->num_backrefs = 0; */ /* XXX: not needed? --apparently not */
|
|
|
|
for (j = 0; match[j]; j++) {
|
|
|
|
/* GNU/POSIX sed does not save more than nine backrefs */
|
2000-11-04 01:17:00 +05:30
|
|
|
if (match[j] == '\\' && match[j+1] == '(' && sed_cmd->num_backrefs <= 9)
|
2000-07-18 01:36:42 +05:30
|
|
|
sed_cmd->num_backrefs++;
|
|
|
|
}
|
|
|
|
|
2000-07-14 01:28:04 +05:30
|
|
|
/* save the replacement string */
|
|
|
|
oldidx = idx+1;
|
2000-07-18 01:36:42 +05:30
|
|
|
idx = index_of_next_unescaped_slash(substr, ++idx);
|
2000-07-14 01:28:04 +05:30
|
|
|
if (idx == -1)
|
|
|
|
fatalError("bad format in substitution expression\n");
|
|
|
|
sed_cmd->replace = strdup_substr(substr, oldidx, idx);
|
|
|
|
|
|
|
|
/* process the flags */
|
|
|
|
while (substr[++idx]) {
|
|
|
|
switch (substr[idx]) {
|
2000-07-15 00:36:30 +05:30
|
|
|
case 'g':
|
2000-11-04 01:17:00 +05:30
|
|
|
sed_cmd->sub_g = 1;
|
2000-07-15 00:36:30 +05:30
|
|
|
break;
|
|
|
|
case 'I':
|
|
|
|
cflags |= REG_ICASE;
|
|
|
|
break;
|
2000-08-22 02:59:20 +05:30
|
|
|
case 'p':
|
2000-11-04 01:17:00 +05:30
|
|
|
sed_cmd->sub_p = 1;
|
2000-08-22 02:59:20 +05:30
|
|
|
break;
|
2000-07-15 00:36:30 +05:30
|
|
|
default:
|
|
|
|
/* any whitespace or semicolon trailing after a s/// is ok */
|
|
|
|
if (strchr("; \t\v\n\r", substr[idx]))
|
|
|
|
goto out;
|
|
|
|
/* else */
|
|
|
|
fatalError("bad option in substitution expression\n");
|
2000-07-14 01:28:04 +05:30
|
|
|
}
|
|
|
|
}
|
2000-07-15 00:36:30 +05:30
|
|
|
|
|
|
|
out:
|
2000-07-18 01:36:42 +05:30
|
|
|
/* compile the match string into a regex */
|
2000-07-14 01:28:04 +05:30
|
|
|
sed_cmd->sub_match = (regex_t *)xmalloc(sizeof(regex_t));
|
|
|
|
xregcomp(sed_cmd->sub_match, match, cflags);
|
|
|
|
free(match);
|
2000-07-15 00:36:30 +05:30
|
|
|
|
|
|
|
return idx;
|
2000-07-14 01:28:04 +05:30
|
|
|
}
|
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
static int parse_edit_cmd(struct sed_cmd *sed_cmd, const char *editstr)
|
2000-07-14 05:30:15 +05:30
|
|
|
{
|
|
|
|
int idx = 0;
|
2000-07-15 04:54:00 +05:30
|
|
|
int slashes_eaten = 0;
|
2000-07-14 05:30:15 +05:30
|
|
|
char *ptr; /* shorthand */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* the string that gets passed to this function should look like this:
|
|
|
|
*
|
|
|
|
* need one of these
|
|
|
|
* |
|
|
|
|
* | this backslash (immediately following the edit command) is mandatory
|
|
|
|
* | |
|
|
|
|
* [aic]\
|
|
|
|
* TEXT1\
|
|
|
|
* TEXT2\
|
|
|
|
* TEXTN
|
|
|
|
*
|
|
|
|
* as soon as we hit a TEXT line that has no trailing '\', we're done.
|
|
|
|
* this means a command like:
|
|
|
|
*
|
|
|
|
* i\
|
|
|
|
* INSERTME
|
|
|
|
*
|
|
|
|
* is a-ok.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (editstr[1] != '\\' && (editstr[2] != '\n' || editstr[2] != '\r'))
|
|
|
|
fatalError("bad format in edit expression\n");
|
|
|
|
|
|
|
|
/* store the edit line text */
|
2000-07-14 06:19:59 +05:30
|
|
|
/* make editline big enough to accomodate the extra '\n' we will tack on
|
|
|
|
* to the end */
|
|
|
|
sed_cmd->editline = xmalloc(strlen(&editstr[3]) + 2);
|
|
|
|
strcpy(sed_cmd->editline, &editstr[3]);
|
2000-07-14 05:30:15 +05:30
|
|
|
ptr = sed_cmd->editline;
|
|
|
|
|
|
|
|
/* now we need to go through * and: s/\\[\r\n]$/\n/g on the edit line */
|
|
|
|
while (ptr[idx]) {
|
|
|
|
while (ptr[idx] != '\\' && (ptr[idx+1] != '\n' || ptr[idx+1] != '\r')) {
|
|
|
|
idx++;
|
|
|
|
if (!ptr[idx]) {
|
2000-07-15 04:54:00 +05:30
|
|
|
goto out;
|
2000-07-14 05:30:15 +05:30
|
|
|
}
|
|
|
|
}
|
|
|
|
/* move the newline over the '\' before it (effectively eats the '\') */
|
|
|
|
memmove(&ptr[idx], &ptr[idx+1], strlen(&ptr[idx+1]));
|
|
|
|
ptr[strlen(ptr)-1] = 0;
|
2000-07-15 04:54:00 +05:30
|
|
|
slashes_eaten++;
|
2000-07-14 05:30:15 +05:30
|
|
|
/* substitue \r for \n if needed */
|
|
|
|
if (ptr[idx] == '\r')
|
|
|
|
ptr[idx] = '\n';
|
|
|
|
}
|
2000-07-15 00:36:30 +05:30
|
|
|
|
2000-07-15 04:54:00 +05:30
|
|
|
out:
|
|
|
|
ptr[idx] = '\n';
|
|
|
|
ptr[idx+1] = 0;
|
|
|
|
|
|
|
|
/* this accounts for discrepancies between the modified string and the
|
|
|
|
* original string passed in to this function */
|
|
|
|
idx += slashes_eaten;
|
|
|
|
|
|
|
|
/* this accounts for the fact that A) we started at index 3, not at index
|
|
|
|
* 0 and B) that we added an extra '\n' at the end (if you think the next
|
|
|
|
* line should read 'idx += 4' remember, arrays are zero-based) */
|
|
|
|
|
|
|
|
idx += 3;
|
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
return idx;
|
2000-07-14 05:30:15 +05:30
|
|
|
}
|
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
static char *parse_cmd_str(struct sed_cmd *sed_cmd, const char *cmdstr)
|
2000-07-11 04:25:51 +05:30
|
|
|
{
|
|
|
|
int idx = 0;
|
|
|
|
|
|
|
|
/* parse the command
|
|
|
|
* format is: [addr][,addr]cmd
|
|
|
|
* |----||-----||-|
|
|
|
|
* part1 part2 part3
|
|
|
|
*/
|
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* first part (if present) is an address: either a number or a /regex/ */
|
|
|
|
if (isdigit(cmdstr[idx]) || cmdstr[idx] == '/')
|
|
|
|
idx = get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match);
|
|
|
|
|
|
|
|
/* second part (if present) will begin with a comma */
|
|
|
|
if (cmdstr[idx] == ',')
|
|
|
|
idx += get_address(&cmdstr[++idx], &sed_cmd->end_line, &sed_cmd->end_match);
|
|
|
|
|
|
|
|
/* last part (mandatory) will be a command */
|
|
|
|
if (cmdstr[idx] == '\0')
|
2000-07-12 22:32:35 +05:30
|
|
|
fatalError("missing command\n");
|
2000-07-14 05:30:15 +05:30
|
|
|
if (!strchr("pdsaic", cmdstr[idx])) /* <-- XXX add new commands here */
|
2000-07-12 22:32:35 +05:30
|
|
|
fatalError("invalid command\n");
|
2000-07-11 04:25:51 +05:30
|
|
|
sed_cmd->cmd = cmdstr[idx];
|
|
|
|
|
2000-07-14 01:28:04 +05:30
|
|
|
/* special-case handling for (s)ubstitution */
|
2000-07-15 00:36:30 +05:30
|
|
|
if (sed_cmd->cmd == 's') {
|
|
|
|
idx += parse_subst_cmd(sed_cmd, &cmdstr[idx]);
|
|
|
|
}
|
2000-07-14 05:30:15 +05:30
|
|
|
/* special-case handling for (a)ppend, (i)nsert, and (c)hange */
|
2000-07-15 00:36:30 +05:30
|
|
|
else if (strchr("aic", cmdstr[idx])) {
|
2000-07-14 05:43:52 +05:30
|
|
|
if (sed_cmd->end_line || sed_cmd->end_match)
|
|
|
|
fatalError("only a beginning address can be specified for edit commands\n");
|
2000-07-15 00:36:30 +05:30
|
|
|
idx += parse_edit_cmd(sed_cmd, &cmdstr[idx]);
|
2000-07-14 05:43:52 +05:30
|
|
|
}
|
2000-07-15 04:54:00 +05:30
|
|
|
/* if it was a single-letter command (such as 'p' or 'd') we need to
|
|
|
|
* increment the index past that command */
|
|
|
|
else
|
|
|
|
idx++;
|
2000-07-15 00:36:30 +05:30
|
|
|
|
|
|
|
/* give back whatever's left over */
|
2000-07-15 04:54:00 +05:30
|
|
|
return (char *)&cmdstr[idx];
|
1999-12-30 03:49:46 +05:30
|
|
|
}
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
static void add_cmd_str(const char *cmdstr)
|
1999-12-30 03:49:46 +05:30
|
|
|
{
|
2000-07-15 00:36:30 +05:30
|
|
|
char *mystr = (char *)cmdstr;
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
do {
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2000-07-15 00:36:30 +05:30
|
|
|
/* trim leading whitespace and semicolons */
|
|
|
|
memmove(mystr, &mystr[strspn(mystr, "; \n\r\t\v")], strlen(mystr));
|
|
|
|
/* if we ate the whole thing, that means there was just trailing
|
2000-07-15 04:54:00 +05:30
|
|
|
* whitespace or a final / no-op semicolon. either way, get out */
|
2000-07-15 00:36:30 +05:30
|
|
|
if (strlen(mystr) == 0)
|
|
|
|
return;
|
|
|
|
/* if this is a comment, jump past it and keep going */
|
|
|
|
if (mystr[0] == '#') {
|
|
|
|
mystr = strpbrk(mystr, ";\n\r");
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
/* grow the array */
|
2000-09-13 08:16:14 +05:30
|
|
|
sed_cmds = xrealloc(sed_cmds, sizeof(struct sed_cmd) * (++ncmds));
|
2000-07-15 00:36:30 +05:30
|
|
|
/* zero new element */
|
|
|
|
memset(&sed_cmds[ncmds-1], 0, sizeof(struct sed_cmd));
|
|
|
|
/* load command string into new array element, get remainder */
|
|
|
|
mystr = parse_cmd_str(&sed_cmds[ncmds-1], mystr);
|
|
|
|
|
2000-07-15 04:54:00 +05:30
|
|
|
} while (mystr && strlen(mystr));
|
1999-11-09 07:17:36 +05:30
|
|
|
}
|
1999-10-28 21:36:25 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
static void load_cmd_file(char *filename)
|
1999-10-28 21:36:25 +05:30
|
|
|
{
|
2000-07-11 04:25:51 +05:30
|
|
|
FILE *cmdfile;
|
|
|
|
char *line;
|
2000-07-14 05:30:15 +05:30
|
|
|
char *nextline;
|
2000-07-11 04:25:51 +05:30
|
|
|
|
2000-09-27 08:13:35 +05:30
|
|
|
cmdfile = xfopen(filename, "r");
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
while ((line = get_line_from_file(cmdfile)) != NULL) {
|
2000-07-14 05:30:15 +05:30
|
|
|
/* if a line ends with '\' it needs the next line appended to it */
|
|
|
|
while (line[strlen(line)-2] == '\\' &&
|
|
|
|
(nextline = get_line_from_file(cmdfile)) != NULL) {
|
2000-09-13 08:16:14 +05:30
|
|
|
line = xrealloc(line, strlen(line) + strlen(nextline) + 1);
|
2000-07-14 05:30:15 +05:30
|
|
|
strcat(line, nextline);
|
2000-07-15 04:54:00 +05:30
|
|
|
free(nextline);
|
2000-07-14 05:30:15 +05:30
|
|
|
}
|
2000-07-15 04:54:00 +05:30
|
|
|
/* eat trailing newline (if any) --if I don't do this, edit commands
|
|
|
|
* (aic) will print an extra newline */
|
|
|
|
if (line[strlen(line)-1] == '\n')
|
|
|
|
line[strlen(line)-1] = 0;
|
2000-07-11 04:25:51 +05:30
|
|
|
add_cmd_str(line);
|
|
|
|
free(line);
|
2000-02-09 01:28:47 +05:30
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
1999-12-30 03:49:46 +05:30
|
|
|
|
2000-07-18 01:36:42 +05:30
|
|
|
static void print_subst_w_backrefs(const char *line, const char *replace, regmatch_t *regmatch)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* go through the replacement string */
|
|
|
|
for (i = 0; replace[i]; i++) {
|
|
|
|
/* if we find a backreference (\1, \2, etc.) print the backref'ed * text */
|
|
|
|
if (replace[i] == '\\' && isdigit(replace[i+1])) {
|
|
|
|
int j;
|
|
|
|
char tmpstr[2];
|
|
|
|
int backref;
|
|
|
|
++i; /* i now indexes the backref number, instead of the leading slash */
|
|
|
|
tmpstr[0] = replace[i];
|
|
|
|
tmpstr[1] = 0;
|
|
|
|
backref = atoi(tmpstr);
|
|
|
|
/* print out the text held in regmatch[backref] */
|
|
|
|
for (j = regmatch[backref].rm_so; j < regmatch[backref].rm_eo; j++)
|
|
|
|
fputc(line[j], stdout);
|
|
|
|
}
|
|
|
|
|
2000-07-26 02:18:44 +05:30
|
|
|
/* if we find a backslash escaped character, print the character */
|
|
|
|
else if (replace[i] == '\\') {
|
|
|
|
++i;
|
|
|
|
fputc(replace[i], stdout);
|
|
|
|
}
|
|
|
|
|
2000-07-18 01:36:42 +05:30
|
|
|
/* if we find an unescaped '&' print out the whole matched text.
|
|
|
|
* fortunately, regmatch[0] contains the indicies to the whole matched
|
|
|
|
* expression (kinda seems like it was designed for just such a
|
|
|
|
* purpose...) */
|
|
|
|
else if (replace[i] == '&' && replace[i-1] != '\\') {
|
|
|
|
int j;
|
|
|
|
for (j = regmatch[0].rm_so; j < regmatch[0].rm_eo; j++)
|
|
|
|
fputc(line[j], stdout);
|
|
|
|
}
|
|
|
|
/* nothing special, just print this char of the replacement string to stdout */
|
|
|
|
else
|
|
|
|
fputc(replace[i], stdout);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2000-07-14 01:31:58 +05:30
|
|
|
static int do_subst_command(const struct sed_cmd *sed_cmd, const char *line)
|
|
|
|
{
|
2000-11-04 01:17:00 +05:30
|
|
|
char *hackline = (char *)line;
|
2000-07-14 01:31:58 +05:30
|
|
|
int altered = 0;
|
2000-11-04 01:17:00 +05:30
|
|
|
regmatch_t *regmatch = NULL;
|
2000-07-14 01:31:58 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* we only proceed if the substitution 'search' expression matches */
|
|
|
|
if (regexec(sed_cmd->sub_match, line, 0, NULL, 0) == REG_NOMATCH)
|
|
|
|
return 0;
|
2000-07-18 01:36:42 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* whaddaya know, it matched. get the number of back references */
|
|
|
|
regmatch = xmalloc(sizeof(regmatch_t) * (sed_cmd->num_backrefs+1));
|
2000-07-18 01:36:42 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* and now, as long as we've got a line to try matching and if we can match
|
|
|
|
* the search string, we make substitutions */
|
|
|
|
while (*hackline && (regexec(sed_cmd->sub_match, hackline,
|
|
|
|
sed_cmd->num_backrefs+1, regmatch, 0) == 0) ) {
|
|
|
|
int i;
|
2000-07-14 01:31:58 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* print everything before the match */
|
|
|
|
for (i = 0; i < regmatch[0].rm_so; i++)
|
|
|
|
fputc(hackline[i], stdout);
|
2000-07-14 01:31:58 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* then print the substitution string */
|
|
|
|
print_subst_w_backrefs(hackline, sed_cmd->replace, regmatch);
|
|
|
|
|
|
|
|
/* advance past the match */
|
|
|
|
hackline += regmatch[0].rm_eo;
|
|
|
|
/* flag that something has changed */
|
|
|
|
altered++;
|
2000-07-18 01:36:42 +05:30
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* if we're not doing this globally, get out now */
|
|
|
|
if (!sed_cmd->sub_g)
|
|
|
|
break;
|
2000-07-14 01:31:58 +05:30
|
|
|
}
|
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* if there's anything left of the line, print it */
|
|
|
|
if (*hackline)
|
|
|
|
fputs(hackline, stdout);
|
|
|
|
|
|
|
|
/* cleanup */
|
|
|
|
free(regmatch);
|
|
|
|
|
2000-07-14 01:31:58 +05:30
|
|
|
return altered;
|
|
|
|
}
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
static int do_sed_command(const struct sed_cmd *sed_cmd, const char *line)
|
|
|
|
{
|
|
|
|
int altered = 0;
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
switch (sed_cmd->cmd) {
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
case 'p':
|
|
|
|
fputs(line, stdout);
|
|
|
|
break;
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
case 'd':
|
|
|
|
altered++;
|
|
|
|
break;
|
|
|
|
|
2000-07-14 01:31:58 +05:30
|
|
|
case 's':
|
2000-11-04 01:17:00 +05:30
|
|
|
|
|
|
|
/*
|
|
|
|
* Some special cases for 's' printing to make it compliant with
|
|
|
|
* GNU sed printing behavior (aka "The -n | s///p Matrix"):
|
|
|
|
*
|
|
|
|
* -n ONLY = never print anything regardless of any successful
|
|
|
|
* substitution
|
|
|
|
*
|
|
|
|
* s///p ONLY = always print successful substitutions, even if
|
|
|
|
* the line is going to be printed anyway (line will be printed
|
|
|
|
* twice).
|
|
|
|
*
|
|
|
|
* -n AND s///p = print ONLY a successful substitution ONE TIME;
|
|
|
|
* no other lines are printed - this is the reason why the 'p'
|
|
|
|
* flag exists in the first place.
|
|
|
|
*/
|
|
|
|
|
2000-11-04 01:32:35 +05:30
|
|
|
/* if the user specified that they didn't want anything printed (i.e. a -n
|
|
|
|
* flag and no 'p' flag after the s///), then there's really no point doing
|
|
|
|
* anything here. */
|
|
|
|
if (be_quiet && !sed_cmd->sub_p)
|
|
|
|
break;
|
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* we print the line once, unless we were told to be quiet */
|
|
|
|
if (!be_quiet)
|
|
|
|
altered = do_subst_command(sed_cmd, line);
|
|
|
|
|
|
|
|
/* we also print the line if we were given the 'p' flag
|
|
|
|
* (this is quite possibly the second printing) */
|
|
|
|
if (sed_cmd->sub_p)
|
|
|
|
altered = do_subst_command(sed_cmd, line);
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
break;
|
2000-07-14 05:30:15 +05:30
|
|
|
|
|
|
|
case 'a':
|
|
|
|
fputs(line, stdout);
|
|
|
|
fputs(sed_cmd->editline, stdout);
|
|
|
|
altered++;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case 'i':
|
|
|
|
fputs(sed_cmd->editline, stdout);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case 'c':
|
|
|
|
fputs(sed_cmd->editline, stdout);
|
|
|
|
altered++;
|
|
|
|
break;
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
|
|
|
|
|
|
|
return altered;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void process_file(FILE *file)
|
|
|
|
{
|
|
|
|
char *line = NULL;
|
|
|
|
static int linenum = 0; /* GNU sed does not restart counting lines at EOF */
|
|
|
|
unsigned int still_in_range = 0;
|
|
|
|
int line_altered;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* go through every line in the file */
|
|
|
|
while ((line = get_line_from_file(file)) != NULL) {
|
|
|
|
|
|
|
|
linenum++;
|
|
|
|
line_altered = 0;
|
|
|
|
|
|
|
|
/* for every line, go through all the commands */
|
|
|
|
for (i = 0; i < ncmds; i++) {
|
|
|
|
|
|
|
|
/* are we acting on a range of matched lines? */
|
|
|
|
if (sed_cmds[i].beg_match && sed_cmds[i].end_match) {
|
|
|
|
if (still_in_range || regexec(sed_cmds[i].beg_match, line, 0, NULL, 0) == 0) {
|
|
|
|
line_altered += do_sed_command(&sed_cmds[i], line);
|
|
|
|
still_in_range = 1;
|
|
|
|
if (regexec(sed_cmds[i].end_match, line, 0, NULL, 0) == 0)
|
|
|
|
still_in_range = 0;
|
2000-03-07 00:50:35 +05:30
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
}
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* are we trying to match a single line? */
|
|
|
|
else if (sed_cmds[i].beg_match) {
|
|
|
|
if (regexec(sed_cmds[i].beg_match, line, 0, NULL, 0) == 0)
|
|
|
|
line_altered += do_sed_command(&sed_cmds[i], line);
|
|
|
|
}
|
2000-03-07 00:50:35 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* are we acting on a range of line numbers? */
|
2000-08-10 05:39:47 +05:30
|
|
|
else if (sed_cmds[i].beg_line > 0 && sed_cmds[i].end_line != 0) {
|
2000-11-04 01:17:00 +05:30
|
|
|
if (linenum >= sed_cmds[i].beg_line &&
|
|
|
|
(sed_cmds[i].end_line == -1 || linenum <= sed_cmds[i].end_line))
|
2000-07-11 04:25:51 +05:30
|
|
|
line_altered += do_sed_command(&sed_cmds[i], line);
|
2000-03-07 00:50:35 +05:30
|
|
|
}
|
2000-02-09 01:28:47 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* are we acting on a specified line number */
|
|
|
|
else if (sed_cmds[i].beg_line > 0) {
|
|
|
|
if (linenum == sed_cmds[i].beg_line)
|
|
|
|
line_altered += do_sed_command(&sed_cmds[i], line);
|
2000-02-09 01:28:47 +05:30
|
|
|
}
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* not acting on matches or line numbers. act on every line */
|
|
|
|
else
|
|
|
|
line_altered += do_sed_command(&sed_cmds[i], line);
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2000-11-04 01:17:00 +05:30
|
|
|
/* we will print the line unless we were told to be quiet or if the
|
|
|
|
* line was altered (via a 'd'elete or 's'ubstitution), in which case
|
|
|
|
* the altered line was already printed */
|
|
|
|
if (!be_quiet && !line_altered)
|
2000-07-11 04:25:51 +05:30
|
|
|
fputs(line, stdout);
|
|
|
|
|
|
|
|
free(line);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
extern int sed_main(int argc, char **argv)
|
|
|
|
{
|
|
|
|
int opt;
|
|
|
|
|
2000-07-25 23:31:20 +05:30
|
|
|
#ifdef BB_FEATURE_CLEAN_UP
|
2000-07-12 03:08:47 +05:30
|
|
|
/* destroy command strings on exit */
|
|
|
|
if (atexit(destroy_cmd_strs) == -1) {
|
|
|
|
perror("sed");
|
|
|
|
exit(1);
|
|
|
|
}
|
2000-07-13 05:05:21 +05:30
|
|
|
#endif
|
2000-07-12 03:08:47 +05:30
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* do normal option parsing */
|
2000-07-21 04:38:40 +05:30
|
|
|
while ((opt = getopt(argc, argv, "hne:f:")) > 0) {
|
2000-07-11 04:25:51 +05:30
|
|
|
switch (opt) {
|
|
|
|
case 'h':
|
|
|
|
usage(sed_usage);
|
1999-12-30 03:49:46 +05:30
|
|
|
break;
|
2000-07-11 04:25:51 +05:30
|
|
|
case 'n':
|
|
|
|
be_quiet++;
|
1999-12-30 03:49:46 +05:30
|
|
|
break;
|
2000-07-11 04:25:51 +05:30
|
|
|
case 'e':
|
|
|
|
add_cmd_str(optarg);
|
|
|
|
break;
|
|
|
|
case 'f':
|
|
|
|
load_cmd_file(optarg);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* if we didn't get a pattern from a -e and no command file was specified,
|
|
|
|
* argv[optind] should be the pattern. no pattern, no worky */
|
|
|
|
if (ncmds == 0) {
|
|
|
|
if (argv[optind] == NULL)
|
|
|
|
usage(sed_usage);
|
|
|
|
else {
|
|
|
|
add_cmd_str(argv[optind]);
|
|
|
|
optind++;
|
|
|
|
}
|
|
|
|
}
|
1999-12-30 03:49:46 +05:30
|
|
|
|
|
|
|
|
2000-07-11 04:25:51 +05:30
|
|
|
/* argv[(optind)..(argc-1)] should be names of file to process. If no
|
|
|
|
* files were specified or '-' was specified, take input from stdin.
|
|
|
|
* Otherwise, we process all the files specified. */
|
|
|
|
if (argv[optind] == NULL || (strcmp(argv[optind], "-") == 0)) {
|
|
|
|
process_file(stdin);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
int i;
|
|
|
|
FILE *file;
|
|
|
|
for (i = optind; i < argc; i++) {
|
|
|
|
file = fopen(argv[i], "r");
|
|
|
|
if (file == NULL) {
|
2000-07-14 07:21:25 +05:30
|
|
|
errorMsg("%s: %s\n", argv[i], strerror(errno));
|
2000-07-11 04:25:51 +05:30
|
|
|
} else {
|
|
|
|
process_file(file);
|
|
|
|
fclose(file);
|
|
|
|
}
|
2000-02-09 01:28:47 +05:30
|
|
|
}
|
1999-11-09 07:17:36 +05:30
|
|
|
}
|
2000-07-11 04:25:51 +05:30
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|